search

crest-lab / crest

CREST - A program for the automated exploration of low-energy molecular chemical space.
https://crest-lab.github.io/crest-docs/
GNU Lesser General Public License v3.0
182 stars 42 forks source link

Crest version 3 using QCG, is it to have a charged solvent? #293

Closed andrewtarzia closed 2 months ago

andrewtarzia commented 2 months ago

Hey all,

I am testing CREST 3 and the QCG algorithm (super impressive by the way!) for solvating charged species. I am currently working with two steps:

  1. adding explicit ions, neutralising the system
  2. and then adding explicit solvent

However, the --chrg command applies to the solute, right? And, so --chrg X is actually nullified when the ions are added? Is there a best way to do this with QCG?

I am currently thinking of using --chrg 0 and --nopreopt, so that the system when the solvent is being added actually will be charge neutral (once the number of solvent is the right number of ions). However, it is not clear to me if xTB (GFN-FF or GFN-2) will then get the charge distributions right. That said, this is a preliminary step to a longer QCG run with the actual neutral solvent and solute (now with the ions) that will include ensemble generation, so maybe it is not super important.

Thank you in advance!

cplett commented 2 months ago

Hi, The QCG can currently only take charges for the solute so adding a charged solvent can be tricky. In such a case, I would recommend neutralizing the solute by adding the ions first, for example, with the aISS docking algorithm in xtb, and then using the neutral structure as solute with a charge of 0. GFN is usually able to assign rather reasonable charges also when ions are included.

andrewtarzia commented 2 months ago

Ahh, I did not realise I could use the aISS outside of this QCG algorithm, thank you, that makes sense!

andrewtarzia commented 2 months ago

I will close this now.